Ontogenetic transition from specialized root hairs to specific root-fungus symbiosis in the dominant Mediterranean seagrass Posidonia oceanica

Terrestrial plants typically take up nutrients through roots or mycorrhizae while freshwater plants additionally utilize leaves. Their nutrient uptake may be enhanced by root hairs whose occurrence is often negatively correlated with mycorrhizal colonization. Seagrasses utilize both leaves and roots and often form root hairs, but seem to be devoid of mycorrhizae. The Mediterranean seagrass Posidonia oceanica is an exception: its adults commonly lack root hairs and regularly form a specific association with a single pleosporalean fungus. Here we show that at two sites in the southern Adriatic, all its seedlings possessed abundant root hairs with peculiar morphology (swollen terminal parts) and anatomy (spirally formed cell walls) as apparent adaptations for better attachment to the substrate and increase of breaking strain. Later on, their roots became colonized by dark septate mycelium while root hairs were reduced. In adults, most of terminal fine roots possessed the specific fungal association while root hairs were absent. These observations indicate for the first time that processes regulating transition from root hairs to root fungal colonization exist also in some seagrasses. This ontogenetic shift in root traits may suggests an involvement of the specific root symbiosis in the nutrient uptake by the dominant Mediterranean seagrass.

structures were recorded: suction cup-like tips and branched finger-like tips (Fig. 1e-j). Upon detachment of the sampled seedlings, some of the root hairs lost their typical tubular character and the spiral-shaped cell wall became apparent (Fig. 1k).
The young seedlings from the ME-33 site possessed no visible DSE fungal colonization. In contrast, the established seedlings were apparently colonized by dark septate (DS) fungal hyphae occurring mostly on the root surface either in the form of single hyphae or as several parallel hyphae (cf. Fig. 2). The observed fungal colonization anatomically and morphologically corresponded to that described in 28,29 and 30 in P. oceanica adults in the NW Mediterranean Sea and the central Adriatic Sea, respectively.
When the site with young seedlings (ME-33) was re-visited in September 2017, no seedlings were found at places where they were firstly spotted in September 2012.
Adult plants. Terminal fine roots of the adult plants (Fig. 2a) typically lacked root hairs but were frequently colonized by DSE fungi. Three forms of root fungal colonization were observed using light microscopy: superficial hyphae, intercellular hyphae in the rhizodermis and the hypodermis and intracellular microsclerotia in the  (Fig. 2b,c). Occasionally, the hyphae formed dense superficial mycelium which spread over the root as loose to dense hyphal mantles (Fig. 2d,e). Neither the rhizodermal cells nor the cortex and the stele contained DS hyphae. When present (sites ME-65 and 66), root hairs of the adult plants formed both the suction cup-like and the finger-like apical structures (Fig. 2f,g), similarly to the seedlings.
Statistical analyses. All root traits evaluated in Subset 1 as well as total fungal colonization and superficial mantle-like colonization in Subset 2 showed significant differences between the sampled plant/substrate types (p < 0.05; Tables 1 and 2). In Subset 1, the young seedlings possessed the highest root hair density but were free of visible fungal colonization (Table 1). In contrast, the adults lacked root hairs but displayed the second highest fungal colonization levels which however did not significantly differ from the highest levels recorded in the established seedlings. The established seedlings displayed considerably lower level of root hairs than the young seedlings (Table 1). In Subset 2, plants from the silty substrate had more than doubled average root hair density than those from the sandy substrate (7.3 ± 3.2 and 2.8 ± 1.1 per 1 mm of the root length, respectively) but this difference was not statistically significant (Table 2). At the same time, plants from the sandy substrate had doubled average total root fungal colonization and more than tripled average colonization by the surface hyphae compared to plants from the silty substrate ( Table 2).
The hair root apical structures were recorded in plants from all four sites. More than one third of root segments from the sandy substrate, 60% from the silty substrate, 87% of root segment of the established seedlings and all but one examined root of the young seedlings possessed at least one root hair with the apical structures. A significantly higher number of root hairs possessing the apical structures as well as the proportion of root hairs with the apical structures were recorded in both sets of seedlings compared to the adults (Table 3).

Discussion
Here we report an intriguing shift in ecologically significant root traits in the dominant Mediterranean seagrass, the endemic P. oceanica, as summarized in Fig. 3. To our knowledge, such a transition from young seedlings possessing dense root hair cover but lacking fungal root colonization to adult plants lacking root hairs but possessing abundant fungal root colonization has never been reported before in any seagrass species. While this shift may resemble negative correlation between root hair production and mycorrhizal colonization typical in terrestrial as well as some aquatic plants 8,15,16 , it rather seems to be connected with the seagrass ontogeny and likely reflects different evolutionary pressures/environmental constraints typical for the three studied ontogenetic phases. These mainly comprise the need 1/ to attach to the substrate in the young seedling phase and 2/ to derive mineral nutrients from the substrate in the adult plant phase.
Ad 1/ Adaptations of submerged plants to the permanent movement of water column in the high-energy coastal/shallow lake areas are crucial for their survival and successful growth 51,52 . In seagrasses, these adaptations emerge already during the seed dispersal and seedling establishment ontogenetic phases 22,53 but seem to be especially important during seedling anchoring 23,54,55 . One such peculiar adaptation in P. oceanica concerns the finest parts of its root system, the root hairs. Despite that these have been reported as infrequent/rare 30,39 to completely absent 24 in adults, recent studies reveal both regular occurrence and high abundance of root hairs in seedlings as well as their crucial importance for seedling establishment, conferring anchorage strength values up to 5.23 N 23 . These "sticky root hairs" or "adhesive root hairs" 23 feature specific apical structures as an apparent adaptation to surf exposure 23 , [this study]. Similar apical swellings have already been observed in non-marine environments, e.g., in liverwort rhizoids colonized by the prominent ericoid mycorrhizal fungus Rhizoscyphus ericae 56 which typically inhabits Ericaceae hair roots worldwide 57 , in root hairs of various Podostemaceae which live on naked rocks in waterfalls ("foot-like broadened tips" 58 ) and in root hairs of aerial roots of Syngonium podophyllum (Araceae) ("hair roots with swollen tips" 59 ). Greater abundance of the adhesive apical structures as well as higher proportion of root hairs with these structures recorded in seedlings compared to adult plants support the assumption that adhesion to the substrate provided by root hairs is significant for survival of P. oceanica seedlings but continuously loses its importance in later ontogenetic phases.    (30)  In addition to the apical swellings, P. oceanica root hairs are strengthened by spirally formed cells walls 30 , [this study], a feature recently reported also by Tomasello et al. 60 . These authors recognized two types of root hairs in P. oceanica growing in NW Sicily (Italy), i.e., "tubular" and "spiral-shaped", where the latter is morphologically identical to the elongated root hairs probably damaged by too high breaking force as reported here and in Vohník et al. 30 (compare Fig. 1h,i, page 46 in Tomasello et al. 60 with Fig. 2o, page 113 in Vohník et al. 30 and Fig. 1k in this study). Similar feature has been recently reported also in some orchids by Bernal et al. 61 ("spiral root hairs"; see Figs 2-6, page 414) and Almeida et al. 62 ("spiral-shaped root hairs"; see Fig. 1g,h, page 124). However, we suggest that instead of representing morphologically distinct root hairs, the "spiral" root hairs represent cracked tubular root hairs, in contrast to, e.g., the shoehorn or spiral root hairs with spoon-like tips reported in Hedera helix (Araliaceae) by Melzer et al. 63 . Indeed, the spirally formed cell walls in P. oceanica root hairs are morphologically  identical with the helical cell walls in S. podophyllum ("helical ribbon-like root hairs"; see Fig. 5, page 513, Fig. 8, page 515 and Fig. 10, page 516 in Yang & Deng 59 ) which "serve as energy-dissipating units to ensure the safe linkage between the root and substrate" (Yang & Deng 59 , page 508). To our knowledge, the significance of root hairs in P. oceanica nutrient uptake has not been rigorously evaluated but seems to be secondary to their role in seedling attachment to the substrate. Indeed, P. oceanica seedlings may rely on the nutrients stored in the enlarged hypocotyl for up to several months after their germination 64 . With progressing time, along with depletion of nutrients from the seed, the roots take over the role in providing majority of nutrition to the plantlets 40 and maintain their dominant position in the nutrition of the adult plants 20 . Our observation that the production of root hairs decreased in older, established seedlings and especially in adults suggests that with progressing time, the newly formed fine roots take over also the anchoring function.
Intriguingly, while the production of root hairs is reduced in established seedlings and especially adults, their terminal fine roots become colonized by DS mycelium thus forming the specific root-fungus symbiosis recently recorded in P. oceanica adults at many localities in the NW Mediterranean and the Central Adriatic 28-30 . Eventually, a major part of the P. oceanica root system may be formed by these often heavily colonized roots. In contrast, this symbiosis was absent in the young seedlings and there are at least three possible explanations: i/ the respective mycobiont(−s) were missing at the exposed microsites colonized by the young seedlings, ii/ young P. oceanica seedlings possess some mechanisms which prevent them from establishing root fungal symbioses as long as they can utilize their own reserves, and iii/ establishment of the specific root-fungus symbiosis is a lengthy process hence this association cannot be detected in early phases of P. oceanica ontogeny. However, our data are too few to allow any definite conclusions with respect to these possibilities.
Ad 2/ Considering that the often heavily colonized terminal fine roots likely represent a site of significant nutrient uptake by P. oceanica, it is tempting to speculate about the role of the specific fungal symbiosis in facilitation of the nutrient influx from the seabed substrate/matte to the seagrass.
Firstly, seagrasses represent a secondary adaptation to the marine environment, i.e., their ancestors were terrestrial plants. Secondly, since the great majority of terrestrial plants rely on mycorrhizae in nutrient uptake from recalcitrant substrates, it is reasonable to suppose that the ancestors of seagrasses were mycorrhizal. Thirdly, the fungal colonization on the root surface in adult P. oceanica plants resembles, at least to some extent, sparse hyphal mantles formed by some contact exploration ectomycorrhizal fungi 30 . On the other hand, a specific plant-fungus interface for nutrient exchange, analogous to arbuscules, the Hartig net or intracellular hyphal coils and pelotons common in terrestrial mycorrhizae, is missing. Fourthly, mycorrhizal colonization is typically negatively correlated with root hair production. On the other hand, also colonization by endophytes with DS hyphae may be negatively correlated with root hair production 65 and these endophytes typically do not take part in nutrient uptake by the host. What is then the eco-physiological functioning of the specific fungal symbiosis in P. oceanica roots? Because the complete lack of regular P. oceanica seed supply and the very complicated growth of the dominant pleosporalean mycobiont in pure culture 29 make manipulative inoculation experiments practically impossible, we can only hypothesize. Perhaps the DS fungi colonizing P. oceanica roots cheat the ancient mechanisms evolved during the long co-evolution of land plants and their root mycobionts 66 by utilizing photosynthetically bound carbon but providing no reciprocal reward to the seagrass (i.e., they behave as biotrophic parasites). Alternatively, this presumably endophytic fungal association is a prelude to evolution of some kind of marine mycorrhizal symbiosis; according to the so-called waiting room hypothesis, some mycorrhizal fungal lineages evolved from former root endophytes, because root endophytism acts as a symbiotic "waiting room" enabling the interacting partners to evolve a tighter mutualistic relationship 67 . To conclude, while there is no solid proof for the involvement of the pleosporalean fungus in the seagrass nutrient uptake, the regular occurrence of the specific root-fungus symbiosis formed by a mycobiont not known from other hosts or environments in the dominant Mediterranean seagrass is intriguing and begs further investigation.
When revisiting the Sutomore sampling site (ME-33), no seedlings were found in September 2017 at the original microsites. Since the seedlings found here in September 2012 occupied rather shallow depths, this observation is in agreement with the results of Piazzi et al. 49 where all seedlings on rocks and gravel settled in the depth of 2 m eventually died. In contrast, Alagna et al. 55 reported successful seedling establishment and survival in 1-3 m depths which probably reflects different hydrodynamic conditions at the studied localities. Hypothetically, the observed die off could be at least in part caused by a lack of compatible DS mycobionts at the exposed microsites colonized by the young seedlings or by the very slow growth of the main DS mycobiont, the Pleosporales sp. MV-2012 29,30 .

Methods
Sampling. Root samples of young seedlings, older and larger already established seedlings and adult plants of  (Table 4, Fig. 4). Additionally, survival of the young seedlings was checked in September 2017. Thirty young seedlings firmly attached to submerged rocks in shallow water (0.5-2 m) under a steep cliff were randomly sampled at the first site (ME-33) while 30 established seedlings growing in deeper water (10-12 m) were randomly sampled at the second site (ME-64). Control adult plants were randomly collected at the third site (ME-36) at the depth of approximately 6 m. For screening of the effect of the substrate type on the root hair density and root fungal colonization, adult plants were randomly collected at a site with silty seabed substrate (ME-65) and at a site with sandy seabed substrate (ME-66). Whole seedlings were carefully detached from the seabed and inserted into 1500 ml plastic flasks with seawater while root bulk samples of adult plants (5 plants at least 5 m apart) were inserted into 50 ml plastic beakers with seawater. Subsequently, seawater was substituted with 30% ethanol and both the flasks and the beakers were transported to the laboratory where they were stored in a fridge (5-8 °C) until processed. Microscopic observations. Thirty approximately 1 cm long root segments were randomly selected from all samples from all sites and screened using an Olympus SZX12 stereomicroscope. Root hair density was calculated as the number of visible root hairs per 1 mm of the root segment length. Semi-thin transversal sections of the root segments prepared using a manual microtome were screened at high magnifications (200-400x) using an Olympus BX60 microscope equipped with Differential Interference Contrast (DIC). Root fungal colonization was estimated separately for superficial hyphae and two types of intraradical colonization structures -microslerotia and intercellular hyphae. Superficial hyphal colonization was calculated as a percentage of the rhizodermal cells with presence of fungi on their surface. Both types of intraradical colonization were estimated as a percentage of the hypodermal cells containing the respective fungal structure. Additionally, total sum of fungal colonization was calculated. Each root segment was divided into sections so that rhizodermal cells and hypodermal cells directly beneath it formed one section. If fungal structures were present either on the surface or in the hypodermis, the section was scored for fungal presence. Fraction of sections with fungal presence out of all sections summed up total fungal colonization.  Table 4. Posidonia oceanica sites sampled in this study. a The numbering of the sites continues from Vohník et al. 28-30 . b This site was re-visited in September 2017 and no seedlings were found.  Table 1. Bar = 10 km. (b) The location of Montenegro (arrowhead) within the Adriatic region and its surroundings. Two-digit country codes follow the ISO 3166-1 alpha-2 standard, XK stands for Kosovo; bar = 100 km.
Some root hairs possessed distinctive apical structures. From each sampling site with recorded root hair presence, thirty root segments possessing root hairs were randomly selected and used for an estimation of the frequency of the apical structures. The number of root hairs possessing apical structures per 1 mm of the root length was counted and the fraction of the root hairs possessing the apical structures was calculated.
Photographic documentation was obtained using an Olympus DP70 camera and the image processing software Quick PHOTO MICRO 2.3 (Promicra Ltd., Czech Republic). Scanning electron microscopy (SEM) was performed using a FEI Quanta 200 scanning electron microscope in the ESEM mode at low temperatures (−12 °C to −3 °C). Photographs were subsequently modified for clarity in Paint.NET (https://www.getpaint.net/index. html) as needed and assembled using the same software. For more details on microscopic techniques and photo processing see Vohník et al. 68 .

Statistical analyses.
For the purpose of statistical analyses, root samples were divided into two Subsets. Subset 1 consisted of young seedlings (sampling site ME-33) and established seedlings (ME-64) together with adult plants from the site ME-36. Subset 2 comprised adult plants from both the silty (ME-65) and sandy (ME-66) substrates at Sveti Stefan (Table 1). For fungal colonization and root hair density values, homogeneity of variances was tested using the Leven's test; normality of the distribution was tested using the Shapiro-Wilk test. Subsequently, in order to determine the influence of the type of the seabed substrate on the root fungal colonization and the density of the root hair cover in Subset 2, the non-parametric Kruskal-Wallis test was carried out. In the case of significant differences between the groups (p < 0.05), the multiple comparison post-hoc Nemenyi test was performed. In Subset 1, no root hair presence was recorded in adult plants and no fungal colonization was observed in the roots of young seedlings. In order to avoid zero inflation of results, we determined the influence of the plant ontogenetic phase on the root fungal colonization and the density of the root hair cover in Subset 1 using Generalized linear model with Gaussian error structure and identity link function.
Frequency of the apical root hair structures and fraction of the root hairs possessing the apical structures were evaluated in a separate analysis using methods identical to those used for Subset 2. In this analysis, all four groups with recorded root hair presence (i.e., young seedlings, established seedlings and adult plants from both the silty and the sandy substrate) were analyzed as a single dataset.
All statistical analyses were carried out using the R software ver. 3.3.2. For calculation of the Nemenyi test, the PMCMR package was employed 69 .
Data availability. The datasets generated during the current study are available from the corresponding author on reasonable request.